Testthat #40
Closed
Testthat #40
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
{pillar} and {vctrs} are the backbone for
customizing tibbles. They are dependencies
of the {tibble} package and therefore
"free" once {tibble} is used as a dependency
package of {STATcubeR}
try this class only with sc_table_saved_list() for now
make sure the objects of class <sc_table_uri> are compatible with sc_table_saved()
- don't import {tibble} since currently,
only {vctrs} and {pillar} is used
- export as.character() for sc_schema_uri
- re-roxygenize
this is now handled as in sc_table(), od_table() and so on
if this package is roxygenized insie of the STAT firewall, the documentation links generated by sc_browse*() will point to the internal server re-roxygenize from the outside TODO: find a way to avoid this in the future. Maybe write a wrapper-function around devtools::document() which temporarily sets the env-var STATCUBER_IN_STAT
another tweak for cli::style_hyperlink(). Hopefully, this will get easier once these features mature
ad some notes that instead of VALUE and VALUESET it is also possible to use uris for COUNT resources in the "measures" parameter of sc_table_custom()
this error was overlooked when the error handling vignette was first written fortunately, the API does a good job of explaining the error in the json body of the response so the error handlerst do not need an upgrade [ci skip]
the sc_table article now showcases the print methods for all the example datasets in german [skip ci]
[skip ci]
add those entries to the metadata. NOTE: columns 5 and 7 are not used in data.csv according the OGD standard but some internale datasets provide these columns and therefore they are imported as the description of the measure/classification
add a patch release since the additional metadata are needed for a deployment NEWS for 0.5.0.1 and 0.5.1 will be merged when 0.5.1 is released
* since json-downloads requir a login, link to the login page * link to the documentation page instead of the manual
- remove @Keywords internal - add documentation for missing params [skip ci]
first attempt to resolve #33. Recodes can now be defined with an additional parameter. However, type-checking is very minimal. TODO: - better error handling when the request is constructed. This way users get quick and useful error messages - at least for semantic errors such as invalid usage of parameters - with this implementation, users will have to make sure that the parameters "recodes" and "dimensions" are consistent. Maybe simplify the usage - The naming sc_recode is almost conflicting with the class sc_recoder. Possibly rename this function - extend the custom tables article to showcase some usecases for recodes and add a short discussion about usage limits - maybe add sc_filter which only allows filter-type recodes and performs stricter type-checks?
showcase the usage of sc_recode in the web documentation.
there are now several checks in place that throw warnings if inputs in sc_table_custom() or sc_recode() are of the wrong schema-type or if other inconsistencies are suspected. See the section called "error handling" in ?sc_table_custom for more details some of those warnings might be replaced with errors in the future part of #33
add a minimum requirement to pillar for the version from 2021-02-22 to make sure the S3 generics format_tbl_footer() is available
don't use the .onLoad hook with base::registerS3method but use the import via NAMESPACE (roxygen) instead [skip ci]
reimplements #36 with a slightly different approach in regards to naming
links to cache files are now clickable and last_modified and cached can will be abbreviated if there is not enough horizontal space
the resouce uris are now displayed similar to sc_schema()
re-sync the roxygen-generated files
add a new parameter `dry_run` to sc_table_custom() which allows to see what request is generated without actually sending it to the API with this option, all type-checks are still applied
update `{cli}` dependency
add a new helper function data_frame() which generates data frame objects and automatically takes care of some common points - to avoid problems with stringsAsFactors, use the vctrs constructor - add the "tbl" class to enable printing the data frames similar to tibbles this makes it possible to skip ssetting strngsAsFactors repedetly and also avoids some places where `class<-` was used previously
use another examplesIf clause to make sure devtoos::check() can be run without an API key
if map is passed as a named vector, this is never supported. therefore, drop the names. This can be useful for subsetting schema objects via single square brackets
dropping those levels causes hierachy information to become unavailable. some new client code now makes use of the hierarchies. The new behavior will generate the fiels as factor columns with unused factor levels.
if json files are downloaded, use cli::progres_along() to show how many json files are still remaining
the (internal) function for parsing open data datasets used to have a parameter to check if levels were dropped. This is no longer necessary since a629296 [ci skip]
don't use registerS3Method for derived classes of data.frame since it does not seem to make any difference
the current palette of colors for sc_schema is only suited for dark editors. For now, add a light theme for the pkgdown reference pages TODO: once the annotation printing is implemented, think about adding a way to customize the color scheme in all STATcubeR. This could also take care of cli_theme_pkgdown() [ci skip]
make sure the links to statcube in the schema vignette open in a new tab TODO: do this globally in R/df_print.R an also use target=_blank for OGD links [ci skip]
assume that the environment where document() runs can be characterized by the NOT_CRAN flag being present (not necessarily FALSE or TRUE). This is to avoid having links to the editing server in the manpages and on pkgdown this will only matter if the docs are built inside the STAT firewall. If at some point, the docs are built via gh-actions, this can be reverted
the following expression will now show all
folders from the catalogue
sc_schema_catalogue() %>%
sc_schema_flatten("FOLDER")
previously, this would only have returned
one entry containig the root folder because
the recursion was stopped as soon as the
appropriate schema type was detected
this change also affects the schema types
GROUP, MEASURE, FIELD and VALUESET in
sc_schema_db() which can also have child nodes
[ci skip]
check the argument against the list of available schema types. the argument is now also coerced via toupper() because the spelling in schema uris uses lowercase
the nace classification in this database was updadet. Reflect this in the example request [ci skip]
[ci skip]
cli_text uses the message channel to generate the visible console outputs this is not what to exprect from a print method wich should always feed into stdout cli_text() is also used in other places of STATcubeR but always wrapped into cli_fmt() which means that output channels do not matter in those circumstances because the outputs are captured to be formatted elsewhere
include another link to github into the DESCRIPTION metadata. this is common practice in most packages on CRAN [ci skip]
first iteration of some unit tests. for now, the goal is just to get a bigger picture about hot/cold code and redundancy TODO: - figure out how to get a reasonable test coverage without an API key - set up a gh-actions job without an API key to make sure that the test run successfully without one (currently, they reqiure a key)
instead of getting the language from the
response headers, this is now an explicit
argument of sc_table_class$new()
the reason for that is supporting {httptest}
which will be used in this branch to
test the package based on cached API responses
the caching in httptest does not store
headers by default. This means that in order
to use the mock responses, the parsers
need to be working even if the headers only
contain a status code and a content type
there are now some additional tests which were inspired by vignettes/sc_last_error.Rmd and check wether the error-types are as expected those tests use httptest. The contents of tests/testthat/statcube* are cached API responses which will be used inside the new tests. the only modification of the caches prior to versioning them was that the support id was removed from some responses.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Just a draft pull request so gh-actions gets triggered by this branch.